Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/HalidSaglam/saglitzdesign-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/halidsaglam/saglitzdesign-mcp/design_review)<a href="https://agentmods.dev/commands/halidsaglam/saglitzdesign-mcp/design_review"><img src="https://agentmods.dev/badge/commands/halidsaglam/saglitzdesign-mcp/design_review/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/halidsaglam/saglitzdesign-mcp/design_review"><img src="https://agentmods.dev/badge/commands/halidsaglam/saglitzdesign-mcp/design_review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00030 | $0.01360 |
| Opus 5 | $0.00015 | $0.00680 |
| Sonnet 5 | $0.00006 | $0.00272 |
| Haiku 4.5 | $0.00003 | $0.00136 |
Grade A, and why
design_review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 67 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Do an expert design review, using the SaglitzDesign method.
You have the SaglitzDesign tools available. Use them — do not design from memory.
Knowledge (judgement):
- get_design_roadmap(project_type) — the phased plan; call FIRST on any build.
- search_design_knowledge(query) / get_design_doc(id) — rules & specs. Ids also readable as saglitzdesign://doc/{id}.
- get_component_guidance(component, platform) — per-component specs + patterns.
- get_design_language(language) — Material 3 / Liquid Glass / iOS / Android / macOS / visionOS / Fluent / web-trends / tokens.
- compare_design_languages(topic, platforms) — how each platform solves one surface, and what NOT to port.
- get_design_examples(query, platform) — curated real-world example screens.
- seo_geo_guide(scope, topic) — SEO & GEO for web.
Generators (real output, not advice — prefer these over inventing values):
- create_design_system(brand_color, vibe, platform) — the one-call foundation: direction card + color + fonts + icons + type scale + elevation + tokens + component list.
- generate_layout_system(preset) — breakpoints, containers, grid, container queries, section rhythm.
- generate_color_system / suggest_font_pairing / suggest_icon_library / generate_type_scale / generate_elevation_system / generate_motion / generate_design_tokens — the individual layers.
- get_component_recipe(component, stack) — production-ready accessible code for button/input/modal/toast/card/switch/tabs/empty-state/list-row/navigation/search/select/table/tooltip/form/pagination/skeleton/badge/breadcrumb.
- fix_contrast(foreground, background, target) — the corrected color value, not just a fail report.
Auditors (deterministic gates — run these before claiming done):
- audit_accessibility(contrast_pairs, tap_targets) — exact WCAG ratios and target sizes.
- design_lint(code) — design & a11y anti-patterns with line numbers.
- audit_design_system(code) — consistency score + value sprawl across the whole codebase.
- audit_project(path) — the same auditors over a real directory instead of a pasted snippet, ranked worst-file-first. Prefer this when you have the source on disk.
- audit_generic_design(code) — the specific defaults generated interfaces reach for (stock indigo/violet/purple gradient, Inter as the only family, the rounded-2xl card recipe, gradient headings, eyebrow-on-every-heading, three or more animate-pulse/shimmer placeholders). Facts about source, not taste. A genuine indigo brand still flags.
- audit_ux_copy(text) — readability, passive voice, jargon, weak CTAs.
- audit_ethical_design(code) — named deceptive-pattern tells in a pasted snippet (confirmshaming decline copy, a pre-checked marketing checkbox, literal scarcity/deadline copy with no live binding, Accept all without Reject all). Facts about source, not a verdict. Pair with get_design_doc("ethical-design").
- audit_apple_ui(path) — iOS/macOS project directory. Reads configuration and Swift. Directory only; no snippet mode.
- audit_android_ui(path) — Android project directory. Reads the manifest, resource XML and Compose. Directory only; no snippet mode.
- measure_screenshot(path, scale, format) — measures a PNG screenshot's real palette, contrast ratios, density and structure, and names a stock-region fact when a significant cluster sits in Tailwind indigo/violet/purple. Use it whenever you have an image file rather than source.
- design_review_checklist(project_type, focus) — the assembled audit checklist.
Sequence
- Identify the project type (website / landing-page / mobile-app / macos-app / dashboard). If it's a URL or running app and a browser tool is available, open it and screenshot mobile + desktop; otherwise review the code/design provided.
- Measure before you opine. If you have the source, run the deterministic auditors first and let their output drive the findings:
- design_lint(code) — anti-patterns with line numbers.
- audit_design_system(code) — is there actually a system, or is every screen re-deciding the basics? This is the finding senior reviewers make and juniors miss.
- audit_accessibility on the real color pairs and target sizes; fix_contrast for each failure.
- audit_ux_copy on the primary headlines, CTAs and error messages.
- audit_ethical_design on consent, decline, urgency and checkout copy when the surface has any of those.
- measure_screenshot(path) — when you are reviewing a screenshot file rather than source, this is the equivalent of the auditors above: real palette, real contrast ratios, real spacing. If a significant cluster sits in Tailwind indigo/violet/purple, the report names that as a stock-region fact — keep it if it is the brand.
- Run design_review_checklist for that type, plus a focused pass where it matters (accessibility, conversion, seo, copywriting).
- Score against get_design_doc("design-critique-scoring") (0–40) with per-heuristic notes.
- Report findings ranked by severity (P0→P3): what's wrong, why (cite the rule/doc, or the measured number), and the concrete fix. Separate "must fix" from "polish". Lead with anything the tools measured — a stated ratio of 2.9:1 or a score of 41/100 ends an argument that an opinion cannot.
- If asked, apply the top fixes and re-run the auditors to show the delta.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago Changed · +5 lines 81bac7a39636
- 8d ago First seen · 62 lines · 30 tokens per session scan A de4ac3ec1282
design_review is a command published in the GitHub repository HalidSaglam/saglitzdesign-mcp (5 stars, last pushed 7d ago), licensed MIT. It adds 30 tokens to every session and 1,360 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
start-here
Find where to start in the collection — name what you're working on and get routed to one command, the two that follow it, and the stages you can skip.
critique-screen
Run all seven visual critiques on a screen and output a prioritised fix list.
explore-options
Run a parallel exploration end to end — frame the decision, build a spread of behaviourally distinct concepts, pressure-test each, and converge with a decision record.
platform-audit
Audit a design for iOS and Android convention compliance — navigation, controls, typography, and platform-specific gaps.
critique-ux
Run a focused UX critique on a screen — affordances, information density, and hierarchy — and output a prioritised fix list.
handoff
Run the full handoff workflow — specs, measurements, assets, states, and a QA checklist — and output a developer-ready package.